R_REDDYX.XYZ
SecurityTypeScript★ 1,437⑂ 189 forksApache-2.0

0sec-labs/0security

SOVEREIGN AI SECURITY AGENTS INFRASTRUCTURE

Open-source infrastructure for deploying sovereign AI security agents. Agents continuously find, verify, and fix vulnerabilities in code, web applications, dependencies, and binaries.

// KEY FEATURES

  • Scans code, web apps, APIs, MCP, and dependencies for vulnerabilities
  • Reverse-engineers binaries via Ghidra and memory fuzzing
  • Creates persistent agents with budget, memory, and schedule
  • Generates patches with verification and SARIF reports

// DEEP DIVE: HOW IT WORKS

0security is a TypeScript CLI tool and platform that runs AI security agents locally or in CI. Agents receive repository context, send it to a chosen LLM (OpenAI, Anthropic, OpenRouter, DeepSeek, or a local OpenAI-compatible server), and execute scans via modules: static code analysis, dependency checking, binary reverse-engineering via Ghidra, attack surface mapping, and prompt-injection testing. Results output as JSON or SARIF, and findings can be reproduced and auto-patched.

Key differentiator is sovereignty and persistence: the agent runs in your infrastructure, only permitted context leaves it, and the agent retains memory, budget, and skills across runs. It includes built-in skills and extensions, plus an MCP server for control from coding agents. The project has disclosed vulnerabilities in the Linux kernel and other open-source projects, proving detection quality.

Requires Node.js 24+ and pnpm to build from source; binaries for macOS (Apple Silicon), Linux (x64/ARM64), and experimental Windows. Needs an API key from at least one LLM provider or a local server. Install via curl script, Docker, or source. Licensed MIT or Apache-2.0.

// USE CASES

Best for: For DevSecOps, security engineers, and teams for continuous code and infrastructure auditing

  • Scan repository on every PR in CI with SARIF report output
  • Run persistent agent for nightly dependency and binary auditing
  • Test LLM agent resilience against prompt injections via built-in campaigns

// QUICK START

curl -fsSL https://raw.githubusercontent.com/0sec-labs/0security/main/install.sh | bash
export OPENAI_API_KEY="your-api-key"
0 scan ./your-repo --runtime api --depth quick --cost-ceiling 2
docker run --rm ghcr.io/0sec-labs/0security:latest --help

Commands are taken from the repository README — check the current version on GitHub before running.

#ai-agents#application-security#binary-analysis#code-review#cybersecurity#llm-agents#mcp#penetration-testing#security-research#security-tools#vulnerability-detection#vulnerability-scanner
Open on GitHub →

New repositories every 30 minutes

REDDYX AI scans GitHub 24/7 and ships the best AI/ML/Web3 projects to Telegram.

Join on Telegram
← Full catalog·Full index

// SIMILAR REPOSITORIES

OPEN SOURCE RED TEAMING TOOLKIT FOR AI★ 4966OPEN-SOURCE SECURITY INFRASTRUCTURE PLATFORM★ 30545REA - REVERSE ENGINEERING WITHOUT BOUNDARIES★ 460891DeepSeek Plugin Framework★ 281341n8n Process Automation★ 210495Claw Code: Fast Code★ 194780

← FULL CATALOG

0sec-labs/0security — Sovereign AI Security Agents…